Set Up Nano Banana Plus 1,000 Prompt Templates

NanoBanana 2 turns Claude Code into a full design studio. Thumbnails, infographics, ads, product shots, icons, diagrams, all from one prompt, plus a free library of 1,000+ copy-paste prompts organized by style. This lesson covers the 2-command install and every tool you unlock.

What You'll Learn

• 2-command install (MCP server plus Claude Code registration)
• 7 built-in tools: generate, edit, restore, icons, patterns, diagrams, storyboards
• 1,000+ free prompts organized across 40+ styles and use cases
• 4K output, text rendering, multi-turn editing, 10 aspect ratios

Why Image Generation Inside Claude Code Matters

Every project needs visuals. Blog posts need featured images. Landing pages need hero graphics. Ad campaigns need creative variations. Presentations need diagrams. And every time you need one, you leave your workflow: open Canva, wait for Midjourney, or message a designer and wait days.

NanoBanana 2 eliminates that. It connects Google's Gemini image models directly to Claude Code as an MCP server. Once installed, Claude has seven image tools built in. You describe what you want, Claude generates it, and the image lands in your project directory. No app switching. No waiting. No separate subscriptions.

And because it runs on Gemini's latest models, the output quality is serious: 4K resolution, accurate text rendering, ten different aspect ratios, and multi-turn editing where you describe changes and Claude updates the image instantly.

Setup (2 Commands)

Step 1: Install the MCP server

The NanoBanana MCP server connects Gemini's image generation directly to Claude Code. Install it globally so Claude can access it from any project.

npm install -g nanobanana-mcp-server

Step 2: Register with Claude Code

Tell Claude Code about the new MCP server. This gives Claude access to all seven image tools automatically.

claude mcp add nanobanana node ~/.gemini/extensions/nanobanana/mcp-server/dist/index.js

Step 3: Set your API key

NanoBanana uses Google's Gemini API for image generation. If you already have a Gemini API key set up, you're good. Otherwise, grab a free key from Google AI Studio and set it as an environment variable.

export GEMINI_API_KEY=your_key_here

Verify the connection: open Claude Code and ask it to generate a simple test image. If an image appears in your project directory, you're connected.

Pro model upgrade: by default, NanoBanana uses Gemini Flash (fast, good quality). For 4K output and better text rendering, set NANOBANANA_MODEL=gemini-3-pro-image-preview in your environment. Higher quality, slightly slower.

The 7 Image Tools

Once connected, Claude Code has seven distinct image tools. Each one handles a different type of visual asset.

1. Generate images. Text-to-image generation with style control. Supports 10 styles (photorealistic, watercolor, oil-painting, sketch, pixel-art, anime, vintage, modern, abstract, minimalist) and 10 aspect ratios (16:9, 1:1, 4:3, 9:16, and more). Generate up to 8 variations at once with different lighting, angles, color palettes, or moods.
2. Edit images. Modify any existing image with natural language. Say "make the background darker" or "remove the text and add a sunset." Claude loads the image, applies your changes, and saves the updated version. Multi-turn editing: keep refining until it's right.
3. Restore images. Repair and enhance old or damaged photos. Fix scratches, improve resolution, correct colors, remove noise. Describe what needs fixing and Claude handles the rest.
4. Generate icons. Create app icons, favicons, and UI elements in multiple sizes (16px to 1024px). Choose flat, skeuomorphic, minimal, or modern styles. Transparent or colored backgrounds. Rounded or sharp corners.
5. Generate patterns. Create seamless patterns and textures for backgrounds, fabrics, and materials. Geometric, organic, abstract, floral, or tech styles. Control density (sparse to dense) and color schemes (mono, duotone, colorful).
6. Generate diagrams. Technical diagrams, flowcharts, architecture diagrams, network maps, wireframes, mind maps, and sequence diagrams. Professional, clean, hand-drawn, or technical styles. Horizontal, vertical, hierarchical, or circular layouts.
7. Generate story sequences. Create 2 to 8 sequential images for narratives, tutorials, timelines, or process flows. Consistent or evolving style across frames. Output as separate files, grid, or comic layout.

The Free Prompt Library (1,000+ Prompts)

Writing good image prompts from scratch is hard. This free website solves that: over 1,000 NanoBanana prompts organized across 40+ use cases and styles. No sign-up required.

youmind.com/nano-banana-pro-prompts

How to use it: browse by use case (thumbnails, infographics, ads, cinematic shots, 3D renders, social posts) or by style (cyberpunk neon, vintage retro, clean minimal, watercolor, anime). Find a prompt you like, copy it, paste it into Claude Code, swap in your subject, and generate. Two minutes from idea to finished image.

Best use cases from the library:

• YouTube thumbnails: eye-catching designs that pull clicks. Multiple styles from bold and graphic to cinematic and moody.
• Infographics: visual breakdowns of complex ideas. Clean layouts with data visualization, process flows, and comparison grids.
• Ad creatives: product shots, lifestyle imagery, and promotional graphics. Styled for social feeds, stories, and display ads.
• Social media posts: scroll-stopping visuals for Instagram, LinkedIn, and X. Quote graphics, announcement posts, and content headers.
• 3D product renders: photorealistic product shots without a photographer. Perfect for ecommerce, pitch decks, and marketing.
• Cinematic shots: movie-quality scene compositions. Great for blog headers, presentation backgrounds, and creative projects.

Prompt editing tip: don't just copy prompts as-is. The best workflow: copy a prompt from the library as your base, swap in your specific subject, then tell Claude to "generate this and then adjust the lighting to be warmer." Multi-turn editing lets you refine without starting over.

Real-World Workflows

1. Blog post featured images. Writing a blog post in Claude Code? When you're done, tell Claude to generate a featured image based on the article topic. It reads the content, picks an appropriate visual concept, and generates the image directly into your blog's image directory. The image matches the content because Claude already understands what you wrote.
2. Landing page hero graphics. Building a landing page? Instead of searching stock photo sites, tell Claude to generate a hero image that matches your brand and message. Generate multiple variations, pick the best one, and it's already in your project folder.
3. Ad campaign creative variations. Need five ad variations for A/B testing? Generate them all in one session. Same product, different styles: photorealistic, minimal, bold graphic, lifestyle, cinematic. Claude generates all five, names them clearly, and saves them to your project directory.
4. Client presentations. Building a pitch deck or proposal? Generate custom diagrams, flowcharts, and visual assets that match your presentation style. Architecture diagrams for technical proposals. Process flows for consulting decks. Product mockups for design pitches.

Prompt Templates

Copy-paste these into Claude Code. Replace bracketed text with your specifics.

YouTube thumbnail:

Generate a YouTube thumbnail for a video about [YOUR TOPIC]. Bold text, high contrast, eye-catching composition. 16:9 aspect ratio. Make it look like something you'd click on.

Blog featured image:

Generate a featured image for this blog post. Read the content and create a visual that captures the main concept. Clean, modern style. 16:9 landscape. No text on the image.

Ad creative variations:

Generate 4 ad creative variations for [YOUR PRODUCT]. Each one should use a different visual style: photorealistic, minimal, bold graphic, and lifestyle. Square format (1:1) for social feeds.

Product shot:

Generate a professional product photo of [YOUR PRODUCT] on a clean background. Studio lighting. Slight shadow. Photorealistic style. High resolution.

Go Deeper

• Prefer a printable version? Download the PDF: https://drive.google.com/file/d/1HvCF2HX6OlzRiLabuOaV2Q7LFfL2JZ5n/view?usp=drivesdk

Want the whole system installed live, with Cooper on the call? Build Your Own AI, the two week sprint runs the first two weeks of August.

Or get a new working tool every Monday inside The AI Operator community on Whop.